← Leaderboard
← Back to Leaderboard
health_bench
#ModelScore
171.2
Claude Opus 4.8
Anthropic
269.8
GPT-5.4
OpenAI
367.1
Gemini 3 Pro
Google DeepMind
464.5
Claude Sonnet 4.6
Anthropic
560.3
DeepSeek V4
DeepSeek
657.8
Qwen3 Max
Alibaba
753.2
Llama 4.1 405B
Meta
Scores are illustrative sample data.
What it measures
health_bench evaluates health and clinical reasoning against rubrics written by practicing physicians, with explicit compliance and safety boundaries baked into the scoring.
Methodology
Physician rubrics — Practicing clinicians set, case by case, what a correct, safe, appropriately-hedged answer looks like.
Safety boundaries — Cross a compliance or safety line and the score drops, however factually right the rest may be.
Dual review — Disputed cases get a second independent pass, with inter-rater agreement reported beside the score.
← Back to Leaderboard